Spatial Transcriptomics · MahmoodLab
hest
A paired histology and spatial transcriptomics dataset linking 1,276 profiles to high-resolution whole slide images.
What hest is
HEST (also designated HEST-1k) is a multimodal biology and medicine dataset produced by MahmoodLab. It brings together spatial gene expression measurements and high-resolution histology data to assist researchers working at the intersection of computational pathology and molecular biology.
The collection addresses the need for co-registered molecular and tissue-level imaging data. It provides 1,276 spatial transcriptomic profiles that are individually linked and spatial-coordinate-aligned to corresponding whole slide images (WSIs), facilitating integrated histological and transcriptomic investigations.
How it was made
The dataset organises spatial transcriptomics profiles alongside paired pathology whole slide images. Each profile is co-registered with its corresponding tissue section image, which is curated to have a spatial resolution with a pixel size smaller than 1.15 µm per pixel.
Key facts
- Profile count
- 1,276 spatial transcriptomic profiles
- Image format
- Whole Slide Images (WSI)
- Image resolution
- Pixel size < 1.15 µm/px
- Primary fields
- Medicine, biology, pathology, spatial transcriptomics
- Preprint reference
- arXiv:2406.16192
What you can use it for
- Histology feature extraction
- Deriving joint visual and transcriptomic representations from aligned microscopic tissue sections and their associated gene expression measurements.
- Image classification
- Training and evaluating classification algorithms across pathology slides informed by spatial transcriptomic phenotypes.
- Pathology image segmentation
- Segmenting morphological structures, tissues, or cellular compartments on whole slide images guided by paired spatial transcriptomic data.
Explore the data
This dataset is gated
Its authors require their terms to be accepted before the data can be read. Accept them on the original dataset page. The platform does not accept terms on your behalf, so the rows are not shown here.Strengths and limits
Good at
- Links and aligns 1,276 individual spatial transcriptomics profiles directly to whole slide images.
- Ensures all paired whole slide images are available at high resolution with pixel dimensions below 1.15 µm/px.
Watch out for
- The dataset contains 1,276 paired profiles, which may restrict large-scale unimodal training compared with purely histopathological archives.
- Image analyses are limited to slides with pixel sizes below 1.15 µm/px.
Questions
Can I look inside hest without downloading it?
Not here: its authors gate it behind terms each user must accept on the original page. Its description and statistics are still shown.
What licence is hest under?
Its authors declare cc-by-nc-sa-4.0. Read the terms on the original page before reusing the data.
What modalities are provided in HEST?
It provides paired spatial transcriptomics profiles and whole slide pathology images that are aligned to one another.
How many samples does the dataset contain?
The dataset contains 1,276 paired spatial transcriptomic and histology profiles.
What is the resolution of the histology images?
Every whole slide image has a pixel size of less than 1.15 µm per pixel.
How to cite it
arXiv:2406.16192
Written from hest's own dataset card. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.